Credit card is a popular payment method and the transaction data keeps track of purchasing activities in people's daily lives. Extracting location of people's activities is an important task in many data mining problems because it may greatly help improve user experience and the service provided to people. Locating people from credit card transactions is equivalent to determining the location of every POS terminal where a payment takes place. This is however not an easy task because the locations of terminals are not usually provided to the credit card issuing companies and only a few terminals can be unambiguously located through map service by providing the merchants' names. In this paper, we propose a system to infer the locations of POS terminals using transaction data and map service. We first construct a transaction graph where the nodes are POS terminals. We then propose a two phase algorithm to find out uncertain and unknown locations of the terminals. In the first phase, we try to eliminate the uncertainty of POS terminals with multiple candidate locations. We show this problem is NP-hard and then give an effective heuristic algorithm to solve it. In the second phase, we compute the locations of unknown POS terminals by propagating the locations of known ones with spatial-temporal constraints. The algorithm is evaluated using a real-world credit card transaction data set and the result is promising for business applications.